Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: DRBES who wrote (80355)	5/20/2002 8:45:26 PM
From: brushwud	Respond to of 275872

(Only two sales since last August against how many for inteL?) WHAT DO YOU GUYS THINK ? Probably it's a bad time to sell AMD if you've been holding since August.

To: DRBES who wrote (80355)	5/20/2002 9:19:16 PM
From: niceguy767	Respond to of 275872

DRBES: "WHAT DO YOU GUYS THINK ?" Perfectly predictable action when a ship begins taking on water...

To: DRBES who wrote (80355)	5/20/2002 11:44:11 PM
From: milo_morai	Read Replies (2) \| Respond to of 275872

<font color=darkred>NW hits a wall on 3.4Ghz, K8 is over 4000+ TMF Author: Wenh99 Number: of 101047 Subject: NW hits a wall on 3.4Ghz, K8 is over 4000+ Date: 5/20/02 10:55 PM NW at 3.4G consumes 119W on .13u process I've got to say that if Intel follows the current Northwood power ramp for released chips, that worst case power will be around 87W (or lower) at 3GHz. It will be interesting to see how Intel handles thermal issues since current overclocks to 3GHz usually involve going over (sometimes way over) 100W worst case TDP. 100W is considered the limit for inexpensive conventional cooling. If Intel is able to scale the Northwood core to 3.4 GHz, it will have to be done without any voltage increases. Generally, power scales with frequency as follows: P = CfV^2 where C= Capacitance, f = frequency and V = voltage. Since power elevates with the square of the voltage, a small bump in voltage has a big impact on power. By frequency increases alone, according to this formula a 3.4GHz Northwood will consume 1.7 times (3.4/2) more power than a 2GHz Northwood. If voltage is increased, and judging from the voltage hikes necessary in Willamette ramping as well as Northwood overclocking reports, it appears very likely that the voltage will have to be raised, probably even before getting to 2.8GHz. 52.4W * 1.7 = 89.1W and this is at Intel's "TDP" which is only 75% of max power, which would then be 89.1/.75 = 119W! Keep in mind that the 2GHz Willamette P4 had a TDP of "only" 71.8W and the core was pretty much maxed. Intel is able to report somewhat lower TDP values than this formula suggests because they allow Vcc_min to decrease as processors increase in clockspeed, but there is obviously a limit when they will have to bump Vcc anyway and I think they are already very near this. Unless they dramatically lower C (and C is proportional to the number of transisitors and their gate sizes) it is hard for me to see how Intel is going to get to 3.4GHz with the current core, much less exceed this speed. I will be surprised if Intel gets to 3GHz this year, and shocked if they reach 3.2. I think a legitimate launch of the 3.4GHz P4 this year is about as likely as Thomas Pabst getting to heaven. If Intel ever gets to 3.4GHz with Northwood, they will likely have to wait for 90nm until the next speed bump. The "hard data" that I am basing my extrapolations on are the performance of existing silicon and basic physics. No handwaving, no magic. It is pretty simple. I think the merit of my work at Tom's Hardware as Senior Editor, as Senior Analyst of Inquest and now my work at our own site, in retrospect, stands up as well as the work of any other analyst in the business. I am now employed for Centaur and tend our site in my spare time. Part of my job at Centaur is to evaluate the relative merits of other MPU designs. What I am stating here on our forum is what I have stated at work. I think it is pretty clear that not only is 3.4GHz a hard ceiling for Northwood, but that Intel will not be reaching this speed this year (if they try to push to this speed, it will be a major strategic mistake unless they can accelerate 90nm parts into Q1 '03). From what I have seen, I do not think Intel will go above 3GHz this year and if they attempt to do so, they will not be successful in producing the fastest parts in quantity. Furthermore, I believe that Intel recognizes that the P4 is a total failure on the mobile front and we can expect to see Banias completely displace the P4M within a year or two from introduction. And a clarification (or two)... I am interpreting "when K8 comes" to be December of this year -- the time when we should see the first Clawhammers become available. If we are looking at the October timeframe that many people still quote, I think the P4 will likely be around 2.8GHz. Because I believe that Intel might be able to reach 3.4GHz with Northwood (at best), I am stating what I believe is the most optimistic forecast for benefits from improvements to Intel process technology. In other words, by the end of the year the best that Intel can hope for from Northwood (and I will stretch this to the time Prescott comes) is to lower power at 3.4GHz to roughly 2GHz Willamette levels. _________________ Van Smith Editor-in-Chief Van's Hardware Journal van@vanshardware.com flickerdown.com As ET stated "OUCH!"